This page describes how we steer thinking and stepping training in the practice. The evidence behind it is on the overview page.
The five exercise sheets are written for patients. Each has three levels. This page names the parameters behind those levels.
Progression parameters
Change one parameter at a time. If you change two at once, you will not know afterwards which one worked.
- Time window for the response
- From 3 seconds down to 1 second. The most effective single parameter and also the riskiest. Shorten it only once the hit rate is stable across two rounds.
- Number of targets
- From 4 to 6 or 8. More targets lengthen the decision time without changing step length.
- Position of the targets
- In front, then to the side, then behind the standing line. Backward and sideways steps demand the largest balance reserve.
- Target distance
- From half a stride to a full stride. Larger distances demand a real weight shift instead of a tapping step.
- Share of no-go signals
- From 20 to 25 or 30 per cent. A higher share makes the task easier because the tendency to respond drops. For more inhibitory load, lower the share to 10 to 15 per cent.
- Rule switching
- Fixed, then announced, then unannounced mid-round. The unannounced switch is the hardest form in this family of exercises.
- Type of secondary task
- Verbal fluency, then counting backwards in threes, then counting backwards while direction is called. Arithmetic loads more heavily than collecting words.
- Load of the secondary task
- Control it through the pace of the prompts. An externally paced task loads more than a self-paced one.
- Surface
- Firm floor, then a mat, then foam. Only for people who already stand safely on that surface without a second task.
- Gaze demand
- Signal in view, then to the side, then requiring a head turn. The head turn additionally challenges the vestibular system.
- Block length
- 20 signals per block, 60 to 90 seconds of rest. Attention tires before the muscles do. Longer blocks then measure fatigue rather than ability.
What to measure
Stepping reaction time is the measure that relates to falls in older people [3]. It is also the measure that improves with step training [6].
Take the median of the individual times rather than the mean. Single very long responses distort the mean strongly, and those outliers are exactly what tired people produce.
Separate two types of error. A step onto a no-go signal is a commission error and measures inhibition. A missed response to a valid signal is an omission and measures attention or speed. Recording both together as one error rate throws away the information.
Calculating dual-task cost
The cost states in per cent how much a performance drops under the dual task [24]:
Cost in per cent = (single task − dual task) ÷ single task × 100
- Calculate the cost for BOTH tasks, the motor one and the cognitive one [24]. Only both together show which task gave way.
- With time-based measures, performance gets worse as the value rises. Reverse the sign there so that a deterioration shows up as a positive cost.
- Measure both single tasks in the same session. A single-task value from last week makes the cost figure unusable.
- Whether dual-task cost predicts future falls is inconsistent across studies [4]. Use it to follow a course over time rather than to make a prognosis.
What to record
One entry per exercise block. Six fields are enough.
- Date, exercise and level
- Number of signals and number of hits
- Median response time in milliseconds
- Commission errors and omissions, kept apart
- Dual-task cost for motor and cognitive task, where measured
- Stops and the reason for them
Also record which single parameter you changed since last time. Without that field a course cannot be interpreted afterwards.
Dosage
From the meta-analysis on dual-task training: 30 minutes three times a week over four weeks for dynamic balance, 50 minutes three times a week over 13 weeks for functional mobility [10]. Both figures assume attendance at 95 per cent of the sessions [10].
For the question of which dose lowers falls, the data in that analysis were not sufficient [10].
Our approach: 20 to 30 minutes, two to three times a week, in blocks of two to three minutes. Target hit rate 75 to 85 per cent. That corridor comes from practice, not from a study.
Exporting data from the device apps
The menu paths below have not been checked against the current app version. They change with updates. Verify them on the device before you rely on them.
BlazePod stores every activity including individual times in the app account. The result of an activity can be displayed and shared. For further work in a spreadsheet you need the file export.
A-Champs ROXs runs its own app on the same principle: select the activity, open the result, share or export.
- Open the app and go to the list of completed activities.
- Select the activity and open the detail view with the individual times.
- Call up the share or export function and choose a spreadsheet format.
- Save the file on the practice computer and merge it with the record there.
Data protection when exporting
- Both apps require an account and store results with the provider. These are services outside Switzerland.
- Do not use real names as player labels. Use a code and keep the link to the person solely in the practice record.
- Transfer the values you want to keep into the practice record. The app is the measuring route, not the archive.
- If you want to work without an account, measure with a stopwatch and a tally. For following one person over time that is enough.
Limits
- The reliability of reaction lights as a measuring device was studied in young athletic adults [25]. We have no comparable work in older people.
- We know of no study showing fewer falls in older people who train with reaction lights. The falls studies used step mats or exercises without a device [6].
- Values from the app are not comparable between devices and set-ups. Keep target distance, height and lighting the same across measurements.
- A dual-task test does not reliably predict future falls [4].
The reference numbers point to the reference list on the overview page. To the reference list